A Unified Contrastive-Generative Framework for Time Series Classification
This work addresses a domain-specific problem for time series analysis, offering a novel hybrid approach that is incremental in combining existing paradigms.
The paper tackles the problem of self-supervised learning for multivariate time series classification by proposing CoGenT, a framework that unifies contrastive and generative paradigms, resulting in up to 59.2% and 14.27% F1 score gains over standalone methods on six datasets.
Self-supervised learning (SSL) for multivariate time series mainly includes two paradigms: contrastive methods that excel at instance discrimination and generative approaches that model data distributions. While effective individually, their complementary potential remains unexplored. We propose a Contrastive Generative Time series framework (CoGenT), the first framework to unify these paradigms through joint contrastive-generative optimization. CoGenT addresses fundamental limitations of both approaches: it overcomes contrastive learning's sensitivity to high intra-class similarity in temporal data while reducing generative methods' dependence on large datasets. We evaluate CoGenT on six diverse time series datasets. The results show consistent improvements, with up to 59.2% and 14.27% F1 gains over standalone SimCLR and MAE, respectively. Our analysis reveals that the hybrid objective preserves discriminative power while acquiring generative robustness. These findings establish a foundation for hybrid SSL in temporal domains. We will release the code shortly.